> ## Documentation Index > Fetch the complete documentation index at: https://docs.together.ai/llms.txt > Use this file to discover all available pages before exploring further. # Preview tokenized data > Preview how sampled rows from a fine-tuning training file will be tokenized before packing. ## OpenAPI ````yaml openapi.yaml POST /fine-tunes/preview openapi: 3.1.0 info: title: Together APIs description: The Together REST API. See https://docs.together.ai for more details. version: 2.0.0 termsOfService: https://www.together.ai/terms-of-service contact: name: Together Support url: https://www.together.ai/contact license: name: MIT url: https://github.com/togethercomputer/openapi/blob/main/LICENSE servers: - url: https://api.together.ai/v1 description: Default environment for APIs - url: https://api-inference.together.ai/v2 description: Optimized environment for inference security: - bearerAuth: [] paths: /fine-tunes/preview: post: tags: - Fine-tuning summary: Preview tokenized data description: >- Preview how sampled rows from a fine-tuning training file will be tokenized before packing. requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/FineTunePreviewRequest' responses: '200': description: Tokenized data preview generated successfully content: application/json: schema: $ref: '#/components/schemas/FineTunePreviewResponse' '400': description: Invalid request or unsupported dataset for preview. content: application/json: schema: oneOf: - $ref: '#/components/schemas/FineTunePreviewMessageError' - $ref: '#/components/schemas/FineTunePreviewDetailError' '401': description: Unauthorized content: application/json: schema: $ref: '#/components/schemas/ErrorData' '404': description: Training file not found. content: application/json: schema: $ref: '#/components/schemas/ErrorData' '500': description: Internal Server Error content: application/json: schema: $ref: '#/components/schemas/FineTunePreviewMessageError' '502': description: Preview service unavailable. content: application/json: schema: $ref: '#/components/schemas/FineTunePreviewMessageError' '503': description: Preview service is at capacity. content: application/json: schema: $ref: '#/components/schemas/FineTunePreviewDetailError' x-codeSamples: - lang: Python label: Together AI SDK (Python) source: | from together import Together client = Together() preview = client.fine_tuning.preview( model="meta-llama/Meta-Llama-3.1-8B-Instruct-Reference", training_file="", ) print(preview) - lang: TypeScript label: Together AI SDK (TypeScript) source: | import Together from "together-ai"; const client = new Together(); const preview = await client.fineTuning.preview({ model: "meta-llama/Meta-Llama-3.1-8B-Instruct-Reference", training_file: "", }); console.log(preview); - lang: JavaScript label: Together AI SDK (JavaScript) source: | import Together from "together-ai"; const client = new Together(); const preview = await client.fineTuning.preview({ model: "meta-llama/Meta-Llama-3.1-8B-Instruct-Reference", training_file: "", }); console.log(preview); components: schemas: FineTunePreviewRequest: type: object description: Request body for previewing tokenized fine-tuning data. required: - model - training_file properties: model: type: string description: >- Name of the base model whose tokenizer and chat template will be used. training_file: type: string description: File-ID of the uploaded JSONL training file to sample for preview. training_method: type: string default: sft enum: - sft description: >- Fine-tuning method to preview. Only supervised fine-tuning is currently supported. train_on_inputs: type: boolean default: false description: >- Whether prompt or user-message tokens should contribute to training loss in the preview. top_k: type: integer default: 5 minimum: 1 maximum: 50 description: >- Maximum number of rows from the start of the training file to tokenize. FineTunePreviewResponse: type: object description: Tokenized preview for sampled rows from a fine-tuning training file. required: - model - dataset_format - max_seq_length - train_on_inputs - rows properties: model: type: string description: Name of the base model used to tokenize the sampled rows. dataset_format: type: string enum: - general - conversation - instruction description: Detected SFT dataset format for the sampled rows. max_seq_length: type: integer description: Maximum sequence length configured for the requested model. train_on_inputs: type: boolean description: Whether prompt or user-message tokens contribute to training loss. rows: type: array description: >- Tokenized preview rows, in the same order as the sampled training file rows. items: $ref: '#/components/schemas/FineTunePreviewRow' FineTunePreviewMessageError: type: object description: >- Error response with a message field returned by the fine-tuning API layer. required: - message properties: message: type: string description: Human-readable error message. FineTunePreviewDetailError: type: object description: >- Error response with a detail field returned by the preview tokenization service. required: - detail properties: detail: type: string description: Human-readable error detail. ErrorData: type: object required: - error properties: error: type: object properties: message: type: string nullable: false type: type: string nullable: false param: type: string nullable: true default: null code: type: string nullable: true default: null required: - type - message FineTunePreviewRow: type: object description: Tokenized representation of one sampled fine-tuning row. required: - input_ids - tokens - labels - trained_spans - num_tokens - num_trained_tokens - truncated properties: input_ids: type: array description: Token IDs produced for the sampled row. items: type: integer description: Token ID from the model tokenizer. tokens: type: array description: Raw token strings produced for the sampled row. items: type: string description: Token text from the model tokenizer. labels: type: array description: Training labels for each token; masked tokens use -100. items: type: integer description: Label value for the token at the same position. trained_spans: type: array description: Half-open token index ranges that contribute to training loss. items: type: array minItems: 2 maxItems: 2 description: Start and end token indexes for one trained span. items: type: integer description: Token index in the row. num_tokens: type: integer description: Total number of tokens in the preview row after truncation. num_trained_tokens: type: integer description: Number of tokens in the row that contribute to training loss. truncated: type: boolean description: Whether the row was truncated to the model maximum sequence length. securitySchemes: bearerAuth: type: http scheme: bearer x-bearer-format: bearer x-default: default ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.